Genetic association studies of candidate genes have historically suffered due to inadequacies of study design, particularly, the selection criteria for the genetic variants. Several methods now exist to characterize the genetic architecture of a gene or region and identify a set of genetic variants or tagging-SNPs (tSNPs). The implementation of such techniques is the basis for effective and comprehensive genetic association studies for complex diseases. A significant issue that remains is the transferability of tSNPs across populations. This has been acknowledged as an important consideration in tSNP selection and has been addressed for geographic ancestry; for example, four ancestral geographic locations are investigated as constituent parts of the HapMap project and the NIEHS SNPs Program. Transferability studies are just beginning. Recent findings of such studies considering geographic ancestry indicate that the Utah tSNPs from HapMap are transferable across multiple European populations. However, one aspect that has been over-looked is the transferability of tSNPs across disease status. Thus far, sequencing panels for variant discovery efforts are neutral, that is, not ascertained based on disease status. The assumption is, therefore, that tSNPs selected from a neutral panel will adequately represent the haplotypes observed in any subgroup that may subsequently be analyzed. Results from studies testing this assumption have the potential for profound implications on association study design. In this application, we investigate the robustness of the assumption that a neutral sequencing panel is adequate for tSNP selection using real data for XRCC2 and colorectal cancer status as well as simulated data. We will characterize the genetic architecture in 4 sequencing panels based on colorectal cancer tumor characteristics and genetic-loading using colorectal cancer cases from high-risk pedigrees and population- based studies. Our results will be compared to those found using neutral panels, as available from the NIEHS SNPs Program. The XRCC2 tSNPs selected and their haplotypes will be analyzed for association with colorectal cancer status and survival in three independent colorectal resources, totaling approximately 2,000 cases and approximately 2,000 controls. In addition, we will investigate transferability across 'disease' ancestry using simulation. We will describe the relative efficacy of the neutral and disease-based panels across simulation and analysis methods. In particular, we will define the limitations of using a neutral panel with an aim to specifically address the potential limitations of using "off-the-shelf" data from the HapMap and the NIEHS SNPs Program. [unreadable] [unreadable] [unreadable]